ARTICLE IN PRESS

Water Research 38 (2004) 475485

A comparative study of Cyperus papyrus and Miscanthidium violaceum-based constructed wetlands for wastewater treatment in a tropical climate
Joseph Kyambaddea,b, Frank Kansiimea, Lena Gumaeliusb, Gunnel Dalhammarb,*
b

Makerere University Institute of Environment and Natural Resources, P.O. Box 7298, Kampala, Uganda Department of Biotechnology, KTH, Royal Institute of Technology, AlbaNova University Centre, S-106 91 Stockholm, Sweden Received 20 March 2003; received in revised form 22 September 2003; accepted 2 October 2003

Abstract The treatment efciencies of constructed wetlands containing Cyperus papyrus L. (papyrus) and Miscanthidium violaceum (K. Schum.) Robyns (synonymous with Miscanthus violaceum (K. Schum) Pilg.) were investigated in a tropical climate (Kampala, Uganda). Papyrus showed higher ammonium-nitrogen and total reactive phosphorus (TRP) removal (75.3% and 83.2%) than Miscanthidium (61.5% and 48.4%) and unplanted controls (27.9% ammoniumnitrogen). No TRP removal was detected in control efuent. Nutrients (N and P) were signicantly higher (po0:015) in papyrus than Miscanthidium plant tissues. Plant uptake and storage was the major factor responsible for N and P removal in treatment line 2 (papyrus) where it contributed 69.5% N and 88.8% P of the total N and P removed. It however accounted for only 15.8% N and 30.7% P of the total N and P removed by treatment line 3 (Miscanthidium violaceum). In addition, papyrus exhibited a signicantly larger (p 0:000) number of adventitious roots than Miscanthidium. Nitrifying bacteria attached to papyrus (2.15 10671.53 105 MPN/g DW) and Miscanthidium roots (1.30 10478.83 102 MPN/g DW) and the corresponding nitrication activities were consistent with this nding. Epiphytic nitriers appeared more important for total nitrication than those in peat or suspended in water. Papyrus root structures provided more microbial attachment sites, sufcient wastewater residence time, trapping and settlement of suspended particles, surface area for pollutant adsorption, uptake, assimilation in plant tissues and oxygen for organic and inorganic matter oxidation in the rhizosphere, accounting for its high treatment efciency. r 2003 Elsevier Ltd. All rights reserved.
Keywords: Constructed wetlands; Cyperus papyrus; Miscanthidium violaceum; Nutrients; Tropical wetlands; Wastewater treatment

1. Introduction In Uganda and other East African countries, the lake fringe swamps and other wetlands are increasingly being used to polish secondary domestic wastewater efuents from adjacent cities [1] as well as surface-run-off and raw sewage from low-income communities [2]. However,
*Corresponding author. Tel.: +46-8-5537-8300; fax: 46-85537-8483. E-mail address: gunnel@biotech.kth.se (G. Dalhammar).

these wetlands are continuously being modied for agricultural activities and infrastructure development thus hampering their treatment efciencies [3]. Few studies have compared the effects of different plant types on specic pollutant removal efciencies [4,5]. Past studies [6,7] described nutrient transformations and faecal coliform removal processes in the two distinct macrophyte zones dominated by Cyperus papyrus (papyrus) and Miscanthidium violaceum (Miscanthidium) in Nakivubo wetland. Azza et al. [2] and Kipkemboi et al. [8] assessed the differential permeability of papyrus

0043-1354/$ - see front matter r 2003 Elsevier Ltd. All rights reserved. doi:10.1016/j.watres.2003.10.008

ARTICLE IN PRESS
476 J. Kyambadde et al. / Water Research 38 (2004) 475485

and Miscanthidium root mats in Nakivubo swamp, suggesting papyrus as having a greater potential to extract nutrients from wastewater due to the thin and loose root mat that allows waterplant interaction than Miscanthidium whose root mat is thick and compact. The present study is part of a larger investigation into management options for enhancing biological nutrient removal from wastewater in Uganda using natural and constructed wetland technology. The aims of this particular pilot study carried out between December, 2001 and August, 2002 were: (a) to compare the extent of nutrient (nitrogen and phosphorus) removal from wastewater by two local plant species dominant in Nakivubo wetland, (b) to investigate the factors responsible for differential nutrient removal rates between the two macrophyte species in pilot constructed wetlands and (c) to assess the technical viability of two macrophyte-based constructed wetlands with respect to wastewater treatment. The need for the study arises from the increased wastewater generation versus constant and/or degrading wastewater treatment facilities due to urbanization and nancial constraints. From the presented data we will extrapolate and speculate on viability of two emergent oating macrophyte species in biological nutrient removal processes and their potential application for wastewater treatment onsite and for small sized communities.

(treatment line 3). CW2 cells were not planted to allow for oxygenation to occur at the airwater interface due to air currents and wind. All cells of treatment line 1 were not planted (control). The planting densities were 38.5 and 61.5 plants/m2 in each constructed wetland planted with papyrus and Miscanthidium, respectively. Plants were suspended in the containers by tying them to a network of wooden pegs/supports. A monitoring program of the constructed wetlands system began 2 months after planting in order to allow the vegetation and bio-lm to establish. The monitoring was carried out for a period of 5 months. 2.2. Loading and water analyses All units were fed with wastewater efuent collected from National Water and Sewerage Corporation (NWSC) Bugolobi Sewage Treatment Works, Kampala at a rate of 0.064 m3/day giving a hydraulic retention time of 5 days. Wastewater ow rates were adjusted manually at the inlet of CW1 units using 24 mm gate valves and measured at the outlet using a measuring cylinder and stopwatch. Physico-chemical parameters (pH, dissolved oxygen (DO), electrical conductivity (EC) and temperature) were measured in situ using portable WissenschaftlichTechnishe Werkstatten (WTW) microprocessor probes and meters. Water sampling was done at the inuent and efuent as well as along the length of the system. Ammonium-nitrogen (NH4-N) was analyzed following direct Nesslerization method [9]. A diazotization method was used to analyze nitrite-nitrogen (NO2-N), while nitrate-nitrogen (NO3-N) and total-nitrogen (TN) were analyzed by Cadmium reduction method and TN after digestion [9]. Total reactive phosphorous (TRP) was analyzed following the ascorbic acid method [9]. All colorimetric determinations were made using a HACH DR-4000 spectrophotometer. 2.3. Measurement of plant biomass, nutrients and root surface area An overview of the rhizomatous habit and the rooting structure of the two plant species is described elsewhere [6,2]. Weekly recruitment of main roots and shoots on all plants was observed during the course of the study. Each plant was examined and newly emerged main roots and newly recruited shoots counted. During examination, plants were carefully removed from the treatment system in order to avoid damage. The weight of plants in each CW unit was taken using a Salter Samson spring balance (Salter Abbey, West Midlands, UK) and recorded prior to the beginning of the study and weekly for 10 weeks during the course of the study. Replicate samples of roots and shoots were collected during the 4th and 5th months of analysis and their dry weight

2. Materials and methods 2.1. Site description The pilot constructed wetland system receiving secondary efuent from Bugolobi Sewage treatment works, Kampala, was set up in December 2001 at Makerere University botanical garden, Kampala, Uganda. The design was a succession through a number of reverse vertical ow containers in which the efuent is discharged at the bottom of the container and displaces the solution from the top of the container, which then ows to the next container (Fig. 1). The CW beds were substrate-free to allow sufcient mixing and optimal contact between wastewater, microorganisms and plant root systems. A split-plot design with treatment line (unplanted control, papyrus and Miscanthidium) as the main plot factors and block (CW1CW6) as subplot factors was used in this study. The three replicate treatment lines 1, 2 and 3 composed of six cells of circular shape which were arranged in series. The respective diameter and depth dimensions were 0.58 and 0.82 m for CW1 (anaerobic), and 0.58 and 0.43 m for CW2CW6. The rst and second cells (CW1 and CW2) for treatment lines 2 and 3 were not planted while CW3CW6 were planted with locally available Cyperus papyrus (treatment line 2) and Miscanthidium violaceum

ARTICLE IN PRESS
J. Kyambadde et al. / Water Research 38 (2004) 475485
Wastewater reservoir

477

1 CW1

CW2

CW3

CW4

CW5

CW6

Effluent

Wastewater reservoir

12mm gate valve

CW1 CW2 CW3 CW4 CW5 CW6 Effluent

Fig. 1. Schematic diagram of pilot constructed wetlands used in this study. Treatment line 1 is unplanted control while lines 2 and 3 were planted with Cyperus papyrus and Miscanthidium violaceum, respectively.

proportions of N and P determined after acid digestion according to Novozamsky et al. [10]. Assuming that the root is in form of a regular cylinder and area of the root ends is negligible, root surface area was calculated as: S 2pRLN rln; 1

2.4. Enumeration of nitrifying bacteria Enumeration of nitrifying bacteria in the peat, water column and on root surfaces was carried out in the rst part of the planted constructed wetland (CW3) using the most probable number (MPN) technique [11]. The media used was according to Schmidt and Belser [12]. The tests were carried out in 10 ml culture tubes with screw caps; the total test volume was 5.0 ml. One milliliter of each 10-fold serial dilution prepared according to Kansiime and Nalubega [6] was inoculated into ve replicate tubes containing 4 ml of sterile test media, mixed and incubated at room temperature in the dark for 36 weeks. Five culture tubes containing only

where S is the root surface area, p is a constant (3.14), R is the average radius of main roots, L is the average length of main roots, N is the average number of main roots, r is the average radius of adventitious roots, l is the average length of adventitious roots and n is the average number of adventitious roots.

ARTICLE IN PRESS
478 J. Kyambadde et al. / Water Research 38 (2004) 475485

test media were included as negative controls. After inoculation, the used root samples were oven dried at 103 C for dry weight determination. 2.5. Nitrication activity Potential nitrication activity was carried out as described by Belser and Mays [13] with slight modications. Briey, 10 g of root samples were incubated aerobically at room temperature in 500 ml Erlenmeyer asks containing 300 ml of test media (1 g neutralized bacteriological peptone (oxoid), 8.5 g NaCl, 0.095 g ammonium sulfate, and 1.06 g sodium chlorate per liter). For water and peat samples, 100 ml was added to 200 ml of test media. In all cases, sodium chlorate was added to inhibit the conversion of nitrite to nitrate [13]. Ammonia oxidation was monitored by analyzing 10 ml samples for nitrite every 45 min using Merck Spectroquant test kit [14] and a Tecan Sunrise Absorbance Microplate Reader. A graph of nitrite concentration versus time was plotted and the slope of the linear regression was calculated to obtain potential nitrication activity. Nitrite concentrations were calculated

from absorbance of samples and standard nitrite (0.2 mg/l) using BeerLamberts law. 2.6. Statistical analyses Statistical analysis was performed with the package MINITAB Release 13.31 for Windows and included analysis of variance (ANOVA), Bartletts and Levenes tests for homogeneity of variance and normality and Tukeys multiple comparisons for differences between means.

3. Results 3.1. Physico-chemical parameters The changes in values of the measured physicochemical variables for the three treatments (unplanted control, papyrus and Miscanthidium planted CWs) are depicted in Table 1. The differences in water pH, conductivity, temperature and dissolved oxygen (DO) between treatments were highly signicant (po0:005 for

Table 1 Mean7standard error values of physico-chemical variables determined in pilot CWs (n 12) Treatment line 1 Variable pH EC (mS/cm) Temp. ( C) DO (mg/l) NH4-N (mg/l) NO2-N (mg/l) NO3-N (mg/l) TN (mg/l) TRP (mg/l) pH EC (mS/cm) Temp. ( C) DO (mg/l) NH4-N (mg/l) NO2-N (mg/l) NO3-N (mg/l) TN (mg/l) TRP (mg/l) pH EC (mS/cm) Temp. ( C) DO (mg/l) NH4-N (mg/l) NO2-N (mg/l) NO3-N (mg/l) TN (mg/l) TRP (mg/l) CW1 7.770.1 796.3716.0 21.67 0.2 0.3570.04 28.970.3 0.2070.08 0.2370.01 58.572.5 15.571.1 7.770.1 796.3716.0 21.670.0 0.3570.04 28.970.3 0.2070.08 0.2370.01 58.572.5 15.571.1 7.770.1 796.3716.0 21.670.0 0.3570.04 28.970.3 0.2070.08 0.2370.01 58.572.5 15.571.1 CW2 8.170.0 739.9 712.1 21.870.1 2.6470.29 26.771.6 0.3770.02 0.3470.01 52.674.1 14.271.2 8.070.0 738710.7 21.670.4 2.5770.42 26.770.9 0.3570.02 0.3570.02 53.073.8 14.571.2 8.370.1 740714.4 22.070.5 2.5270.32 26.771.5 0.3670.01 0.3470.01 52.073.9 14.870.4 CW3 8.370.1 695.374.3 21.770.1 2.6970.58 26.771.2 0.0670.00 0.3070.01 48.672.6 14.571.5 6.870.0 457.37 13.5 22.270.2 0.9870.24 16.171.0 0.0970.03 0.1670.01 26.372.6 6.370.6 7.170.0 510.876.5 22.270.3 0.5870.05 16.970.5 0.0770.02 0.1470.01 30.572.5 10.471.0 CW4 8.470.1 675.0729.2 21.570.1 3.5870.98 26.271.7 0.0670.01 0.2870.01 51.274.6 16.971.4 6.870.0 386.178.3 22.070.3 0.5370.05 11.370.2 0.0670.01 0.1370.00 25.271.2 4.870.2 7.170.0 470.8712.1 22.170.2 0.570.08 15.370.7 0.0770.02 0.1370.01 24.471.2 10.171.1 CW5 8.370.1 642.8727.9 21.370.1 3.8271.1 23.071.4 0.0970.02 0.2970.00 44.672.1 16.071.3 6.670.0 325.674.9 21.870.2 0.5870.08 8.770.6 0.0270.00 0.1070.01 15.571.2 3.370.3 7.070.0 416.5718.9 22.070.1 0.770.11 12.471.1 0.0570.01 0.1170.01 19.471.2 8.771.1 CW6 8.270.1 631.5733.1 21.470.1 2.3070.68 20.870.3 0.0770.01 0.2870.00 43.572.4 17.171.3 6.770.0 29872.7 21.770.1 0.7670.13 7.171.1 0.0270.00 0.0970.03 16.171.1 2.671.1 7.170.0 407.3729.7 22.270.1 0.9170.2 11.171.7 0.0470.00 0.0770.01 17.972.2 8.071.1

ARTICLE IN PRESS
J. Kyambadde et al. / Water Research 38 (2004) 475485 479

all variables). Water pH was signicantly lower in papyrus (p 0:001) than Miscanthidium (p 0:006) CWs compared to the controls. Similarly conductivity was lower in papyrus (p 0:02) than Miscanthidium (p 0:03) CWs, whereas DO was lower in Miscanthidium (p 0:003) than papyrus (p 0:004) CWs. Multiple comparisons detected higher water temperature in Miscanthidium (p 0:001) than papyrus (p 0:054). The longitudinal variation of the average concentrations of ammonium-nitrogen (NH4-N), nitrite-nitrogen (NO2-N), nitrate-nitrogen (NO3-N), total-nitrogen (TN) and total reactive phosphorus (TRP) is presented in Table 1. The differences in concentrations of NH4-N, NO3-N, TN and TRP between the three treatments were highly signicant (po0:004 for all variables). In contrast no signicant differences (p 0:281) in NO2-N concentrations were detected between treatments. In all treatments, nitrite-nitrogen and nitrate-nitrogen were detected but only in small concentrations. Multiple comparisons showed signicantly higher NH4-N removal in papyrus (p 0:02) than Miscanthidium (p 0:026) CWs compared to unplanted controls. On the other hand, NO3-N was signicantly higher in unplanted controls than papyrus (p 0:027) and

Miscanthidium (p 0:02) CWs. However, multiple comparisons detected signicantly high TN removal in both papyrus (p 0:018) and Miscanthidium (p 0:017) CWs that was nearly equal. Similarly, the longitudinal TRP removal was signicantly higher in papyrus (p 0:003) than Miscanthidium (p 0:004) CWs in comparison with the unplanted controls. 3.2. Recruitment of roots, shoots and weight increment Results of recruitment of main roots and shoots are shown in Fig. 2. The root recruitment rate per CW unit was 77 and 32 roots per week for papyrus and Miscanthidium, respectively. One-way ANOVA did not detect signicant differences in root recruitment rates between the two plant species (p 0:072) despite papyrus having a two-fold higher rate. The trend of weight increment showed a similar pattern to that of root recruitment though it did not tally with shoot recruitment (Figs. 2 and 3). One-way ANOVA showed that weight increment was signicantly different (p 0:014) between the two plant species. The average plant weight increment was 0.228 and 0.039 kg per CW unit for papyrus and Miscanthidium, respectively.

No. of new main roots recruited per CW unit

500 450 400 350 300 250 200 150 100 50 0 0 1 2 3 4 5 6 7 Period (weeks) Papyrus Miscanthidium

No. of new shoots per CW unit

60 50 40 30 20 10 0 wk0 wk1 wk2 wk3 wk4 wk5 wk6 wk7 wk8 wk9 wk10 Period (weeks) Papyrus Miscanthidium

Fig. 2. Average weekly recruitment of roots and shoots per CW planted with Cyperus papyrus and Miscanthidium violaceum.

ARTICLE IN PRESS
480 J. Kyambadde et al. / Water Research 38 (2004) 475485

Though shoot recruitment rates were higher in papyrus than Miscanthidium CWs (5 and 2 shoots per week), respectively, one-way ANOVA did not detect signicant differences between the two plant species (p 0:14).

canthidium. The mean radii of the main root of both plant species did not vary signicantly (p 0:196). In contrast, the average length of the main root was signicantly greater (p 0:03) in Miscanthidium than in papyrus.

3.3. Estimation of root surface area of plants 3.4. Nitrifying bacteria Results of root surface area determinations are shown in Table 2. Papyrus exhibited a signicantly larger number of adventitious roots (p 0:000) than MisWeight increment (kg) per CW unit 3 2.5 2 1.5 1 0.5 0 wk0 wk1 wk2 wk3 wk4 wk5 wk6 wk7 wk8 wk9 wk10 Period (weeks) Papyrus Miscanthidium y = 0.0015x + 0.0165x 2 R = 0.914
2

The distribution of nitrifying bacteria in the different layers of the constructed wetlands is depicted in Fig. 4.

y = 0.0225x - 0.0372x 2 R = 0.9834

Fig. 3. Average weekly increment of plant weight per CW planted with Cyperus papyrus and Miscanthidium violaceum.

Table 2 Parameter measurements used to estimate root surface area of Cyperus papyrus and Miscanthidium violaceum (n 3) Parameter Mean length of main root (cm) Mean length of adventitious roots (cm) Mean number of adventitious roots/main root Mean radius of main root (cm) Mean radius of adventitious roots (cm) Average root surface area (cm2)a
a

Papyrus 24.6570.10 2.2070.03 27777 0.2170.01 0.04670.00 208.6

Miscanthidium 29.5570.74 1.4870.03 5373 0.2370.02 0.06070.01 72.2

Calculated using Eq. (1).

Log number (MPN/100 ml or gDW)

1.00E+08 1.00E +07 1.00E+06 1.00E+05 1.00E+04 1.00E+03 1.00E+02 1.00E+01 1.00E+00 Water Peat Papyrus roots Sample Miscanthidium roots

Fig. 4. Average spatial distribution of nitrifying bacteria in water, peat and on roots of Cyperus papyrus and Miscanthidium violaceum in a pilot constructed wetland. Bars indicate standard error of the mean (n 3).

ARTICLE IN PRESS
J. Kyambadde et al. / Water Research 38 (2004) 475485 481

Numbers of ammonia oxidizing bacteria followed the pattern peat>papyrus>Miscanthidium>water column. The mean numbers of nitrifying bacteria in the papyrus root mat (2.15 10671.53 105 MPN/g DW) were signicantly higher (p 0:000) than those associated with the Miscanthidium root mat (1.30 104 78.83 102 MPN/g DW).

and Miscanthidium were 0.0310, 0.0314, 0.147 and 0.146 mg NO2/l/h. 3.6. Nitrogen and phosphorus content in plant tissues Results of nutrient content in plant tissues are depicted in Fig. 6. Concentrations of N were signicantly higher than P in both papyrus (p 0:037) and Miscanthidium (p 0:004) tissues. In addition, N levels were signicantly higher (p 0:014) in papyrus than in Miscanthidium. Similarly, P levels were signicantly higher (p 0:04) in papyrus than in Miscanthidium.

3.5. Potential nitrication activity Results of potential microbial nitrication activities for water, peat and root mat samples of papyrus and Miscanthidium are depicted in Figs. 5a and b. Higher nitrication potential was found to be associated with papyrus roots than Miscanthidium, water or peat samples (Fig. 5a). The root mat samples of the two plant species showed similar nitrication capacities for the rst 3 h. However, papyrus surpassed to attain a maximum nitrite production of 0.8 mg/l in the next 1.5 h. In contrast, Miscanthidium root samples attained a maximum of 0.514 mg/l in the rst 3 h before it leveled off (Fig. 5b). Similarly, high activity was detected in the peat although it was not signicantly different from that of water (p 0:053). The activity for peat samples reached a maximum of 0.213 mg nitrate-N/l in 7 h as opposed to a maximum of 0.183 mg nitrate-N/l in the water column samples after 5.5 h (Fig. 5b). The slopes of linear regression for water, peat and roots of papyrus

N and P content (g/kg dry weight)

20 18 16 14 12 10 8 6 4 2 0
t sh oo t sh oo t ro o ro ot

Nitrogen Phosphorus

Pa py ru s

is ca nt hi di um

Pa py ru s

Plant tissue

Fig. 6. Nitrogen and phosphorus content in root and shoot tissues of Cyperus papyrus and Miscanthidium violaceum during the 4th and 5th months of analysis. Bars indicate standard error of the mean (n 3).

Nitrite-N concentration (mg/l)

Nitrite-N concentration (mg/l)

0.25 0.2 0.15 0.1 0.05 0 0 2 4 6 Time (h) Nitrite-water Nitrite-Peat 8 10 12

0.25 0.2 0.15 0.1 0.05 0 0 1 2 3 4

M is ca nt hi di um
y = 0.0309x R2 = 0.9886 y = 0.0299x R2 = 0.9816 5

Time (h) Nitrite-water Nitrite-Peat

0. 9 0. 8 0. 7 0. 6 0. 5 0. 4 0. 3 0. 2 0. 1 0 0 100 200 300 400 500 600 700

Nitrite-N concentration (mg/l)

Nitrite-N concentration (mg/l)

0.9 0.8 0.7 0.6 0.5 0.4 0.3 0.2 0.1 0 0 50 100 150 200

y = 0.0027x 2 R = 0.9649

y = 0.0025x 2 R = 0.9798

250

300

350

(a)

Time (min) Papyrus Miscanthidium

(b)
Papyrus

Time (min) Miscanthidium

Fig. 5. (a) Potential nitrication activities of water, peat, papyrus and Miscanthidium root samples. (b) Linear part of nitrication rates and the corresponding root square values.

ARTICLE IN PRESS
482 J. Kyambadde et al. / Water Research 38 (2004) 475485

4. Discussion Our qualitative observations of several quantied water quality parameters suggest a high potential of both plant species in biological nutrient removal processes. In both planted CWs, water pH was within the allowable range of 49.5 for the survival of most bacteria [15]. The water pH value was slightly acidic (pH 6.66.8) in CWs planted with papyrus compared to Miscanthidium (pH 7.07.1; Table 1) probably due to decomposition products of wastewater components [16] trapped in the papyrus root mat as well as decomposing plant materials. Our investigation indicated a two-fold higher root recruitment rate for papyrus with many adventitious root structures than Miscanthidium (Fig. 2; Table 2), which effectively trapped organic suspended particles in feed wastewater and dead plant material. Contrary to eld observations, the water pH value in Miscanthidium CWs was higher than pH 5.0 values reported for Miscanthidium root mats in Nakivubo wetland [17] and other wetlands on the northern shores of Lake Victoria [8]. The signicant difference in water temperature between the three treatments was probably due to the small scale of the CWs. Unplanted CWs showed higher DO concentrations than the corresponding planted wetland units and this was attributed to algal photosynthesis. Effective nitrication has been reported in systems with DO content of only 0.5 mg/l [18]. In this study, the concentrations of nitrite and nitrate in the unplanted control systems were low despite DO concentrations higher than 2.5 mg/l. This is attributed to denitrication and lack of sufcient attachment sites and subsequent ow through of nitricants. In contrast, concentrations of DO decreased in all vegetated CWs due to aerobic decomposition of plant materials, nitrication and minimal surface aeration resulting from vegetation coverage. The low nitrate levels in planted CWs can be explained by heterotrophic competition with nitricants for oxygen [19], denitrication and plant uptake. The values of electrical conductivity and the concentrations of NH4-N, NO2-N, NO3-N, TN and TRP were generally reduced in all CWs. Reduction in these quantied water quality parameters was well demonstrated in planted CWs and was generally higher in papyrus than Miscanthidium CWs. TRP efuent concentrations below the Uganda regulatory discharge limit of 10 mg/l [20] were obtained in both planted CWs, with papyrus showing much lower concentrations of up to 2.6 mg/l. However, the efuent NH4-N concentration below the national discharge limit of 10 mg/l could only be obtained in papyrus CWs. The longitudinal NH4-N and TRP removal rates of 75.3% and 83.2% in papyrus and 61.5% and 48.4% for Miscanthidium-based CWs were achieved compared to only 27.9% NH4-N removal and 10.3% TRP accumulation in unplanted controls. Thus, both macrophytes

demonstrated a positive inuence on nutrient removal processes. The removal values for ammonium (75.3%) and phosphorus (83.2%) obtained in the papyrus systems were high compared to values reported by Lizhibowa [21] (70% NH4-N) in bucket experiments at Kiriinya [22] (o20% for both NH4-N and TRP) in CW pilot units at Kiriinya. Ammonium-nitrogen was removed through plant uptake and microbial nitrication/denitrication processes since temperature and pH were within the range that could support both nitrication and denitrication processes [18]. Ammonia volatilisation could not have occurred because the pH in the system never rose above pH 9 [23]. In aquatic systems, phosphorus removal is mainly via biological uptake by bacteria, phytoplankton and plants [24,25] and adsorption and precipitation by a substratum surface containing free Ca2+, Fe2+, Mg2+ and Al3+ ions [26,27]. Even though epiphytic uptake was not quantied, it could not account for the differences observed in the treatment systems since they were all exposed to similar conditions. As our design was substrate-free, immobilization and soil sorption could not inuence the removal processes. However, the possibility of phosphorus precipitation, adsorption of soluble phosphorus to the roots systems and suspended solids settling at the bottom of treatment CWs could not be ruled out. Mass balance calculations showed plant uptake and storage contribution of 69.5% N and 88.8% P of the total N and P removed by papyrus treatment line and 15.8% N and 30.7% P of the total N and P removed by Miscanthidium violaceum treatment line. Therefore active uptake and incorporation into plant tissue was the major factor responsible for the observed phosphate removal rates in CWs planted with Cyperus papyrus. Other processes such as nitricationdenitrication and adsorption of soluble phosphorus to roots and peat were more important for N and P removal in CWs planted with Miscanthidium violaceum. The differences in the structure and recruitment of roots by the two macrophytes depicted important consequences for the degradation of wastewater components and uptake of nutrients. Indeed, this was supported by the difference in root surface area (Papyrus, 208.6 cm2 per main root compared to Miscanthidium, 72.2 cm2) and recruitment rates between the two macrophytes. Papyrus exhibited a two-fold higher root recruitment rate than Miscanthidium. Similarly, the high number of adventitious root structures conferred papyrus a three-fold larger root surface area than Miscanthidium. Wetland plants are reported to transfer photosynthetic oxygen to the rhizosphere thus boosting oxygen concentration in the water column [28,29]. Assuming the oxygen transfer rate of papyrus to be half the maximum transfer rate (12 g/m2/day) reported for Phragmites australis [30], papyrus CWs provided

ARTICLE IN PRESS
J. Kyambadde et al. / Water Research 38 (2004) 475485 483

6.3 g O2/m2/day compared to the oxygen demand of 5.7 g/day. It is probable that the high root numbers of papyrus coupled with its larger root surface area due to adventitious root structures provided more oxygen to the rhizosphere of root mat, reducing competition between heterotrophs and nitricants in papyrus CWs. Differences in root structures and numbers inuenced periphyton attachment and material transformations in wetland systems [31,32]. In addition to providing attachment sites and diffusible oxygen to the bacteria, root mats increase wastewater residence time and retention of suspended organic particles, which upon degradation avail nutrients to bacteria and plants. In fact, more nitrifying bacteria were associated with papyrus roots (2.15 106 71.53 105 MPN/g DW) compared to Miscanthidium roots (1.30 10478.83 102 MPN/g DW) and water column (2.93 10378.83 101 MPN/100 ml), suggesting more attachment sites and easy accessibility of nutrients. The numbers of nitrifying organisms on papyrus roots and in water column were within the range found in Nakivubo wetland [6]. The numbers of nitrifying bacteria enumerated in the peat (1.67 10773.33 105 MPN/100 ml) and epiphyton were comparably higher than those reported by . Korner [33]. Nitrifying bacteria in the peat might have been inactive in the treatment system and may not have contributed much to the nitrication process but only became active when conditions were favorable in our laboratory. Indeed, this was consistent with the results of potential nitrication activities of plant roots, water and peat samples. Higher activity was associated with plant roots particularly papyrus roots (Fig. 5a) and tallied with MPN numbers (Fig. 4) except for the peat samples. The larger surface area of papyrus root structures attracted more nitrifying bacteria by providing attachment sites, diffusible oxygen from the aerial parts, and substrates from trapped decomposing wastewater components. Nitriers in the epiphyton have been estimated to have an equally high inuence on total system nitrication as those in the sediment although the peat is more important for denitrication [33]. Results from this study showed a 5-fold higher nitrication activity in the epiphyton than in the water column and sediment suggesting epiphytic nitriers to be more important for total system nitrication. One possible explanation is that the nitrifying bacteria in the peat were not so active due to competition for the limited oxygen with heterotrophs. However, their re-growth and activity was supported in our culture tubes where conditions were ambient. Thus, suspended nitrifying organisms in the constructed wetlands positively inuenced total system nitrication bearing in mind the long retention time [34] and fairly high numbers (2.93 103+8.83 101 MPN/ 100 ml) in the water column.

Microbial attachment and root development seemed to have inuenced positively differential nutrient uptake by the two macrophytes. More shoots were developed by papyrus with a corresponding increase in plant fresh weight, a possible indicator that it assimilated more nutrients than Miscanthidium. In fact, analysis of phosphorus and nitrogen content in roots and shoots of both macrophytes (Fig. 6) indicated higher concentrations of the two nutrient variables in papyrus than Miscanthidium plant tissues.

5. Conclusion To evaluate the potential application of a macrophyte in wastewater treatment constructed wetlands, knowledge of structural development and recruitment rates of roots and the general growth rate of the macrophyte in question is crucial. This inuences plantmicroorganismswastewater interactions by providing microbial attachment sites, sufcient wastewater residence time, trapping and settlement of suspended wastewater components as a result of resistance to hydraulic ow, surface area for pollutant adsorption, uptake and storage in plant tissues, and diffusion of oxygen from aerial parts to the rhizosphere. In this study, we observed high wastewater treatment efciencies associated with planted CWs compared to unplanted controls. Papyrus CWs showed markedly higher nutrient removal efciencies, with higher N and P levels in plant tissues in comparison to Miscanthidium. Similarly, more shoots were developed by papyrus than Miscanthidium possibly indicating differences in nutrient uptake. This study showed that epiphytic nitriers were more important for total nitrication than those in peat or suspended in water and also demonstrated that root development and its association with microorganisms is responsible for the high wastewater treatment potential of papyrus that was reported in past studies. At present, we can speculate on the potential application of papyrus for wastewater treatment especially for small-sized communities and onsite treatment and attribute the observed differential nutrient removal rates to a combination of factors stated above, but these remain the topics of our ongoing research activities.

Acknowledgements This study received nancial support from the Swedish International Development Cooperation Agency (Sida)/Department of Research Cooperation (SAREC) under the wastewater treatment component of the East African Regional Programme and Research Network for Biotechnology, Biosafety and Biotechnology Policy Development (BIO-EARN). Authors are

ARTICLE IN PRESS
484 J. Kyambadde et al. / Water Research 38 (2004) 475485 [14] Gumaelius L, Smith EH, Dalhammar G. Potential biomarker for denitrication of wastewaters: effects of process variables and cadmium toxicity. Water Res 1996;30(12):302531. [15] Kadlec RL, Knight RL. Treatment wetlands. Boca Raton, FL, USA: CRC Press-Lewis Publishers; 1996. [16] Verhoeven JTA. Nutrient dynamics in minerotrophic peat mires. Aquat Bot 1986;25(2):11738. [17] Kansiime F, Nalubega M, van Bruggen JJA, Lijklema L. Variation of water quality in the Nakivubo wetland, Kampala-Uganda. Proceedings of Seventh International Conference on Wetland Systems for Water Pollution Control, November 1116, Florida, 2000. p. 72734. [18] Cloete TE, Muyima NYO. Microbial community analysis: the key to the design of biological wastewater treatment systems. IAWQ Scientic and Technical Report No. 5. Cambridge: Cambridge University Press; 1997. 10pp. [19] Focht DD, Verstraete W. Biochemical ecology of nitrication and denitrication. Adv Microb Ecol 1977;1:135214. [20] NEMA, 1999. The national environment (standards for discharge of efuent into water or on land) regulations, Uganda. [21] Lizhibowa M. Effects of pre-settled waste water on the growth of Cyperus papyrus,its performance on nutrient removal in Uganda. M.Sc. thesis, IHE, Delft, The Netherlands, 1995. 173pp. [22] Okurut TO, Rijs GBJ, van Bruggen JJA. Design and performance of experimental constructed wetlands in Uganda, planted with Cyperus papyrus and Phragmites mauritianus. Water Sci Technol 1999;40(3):26571. [23] Eighmy TT, Bishop PL. Distribution and role of bacterial nitrifying populations in nitrogen removal in aquatic treatment systems. Water Res 1989;23(8):94755. [24] Schreijer M, Kampf R, Toet S, Verhoeven J. The use of constructed wetlands to upgrade treated sewage efuents before discharge to natural surface water in Texel island, The Netherlandspilot study. Water Sci Technol 1997; 35(5):2317. [25] Hunt PG, Poach ME. State of the art for animal wastewater treatment in constructed wetlands. Proceedings of the Seventh International Conference on Wetland Systems for Water Pollution and Control. International Water Association 2000;1:2936. [26] Johansson L. Use of light expanded clay aggregates (Leca) for removal of phosphorus from wastewater. Water Sci Technol 1997;35(5):8793. [27] Reddy KR, DAngelo EM. Biogeochemical indicators to evaluate pollutant efciency in constructed wetlands. Water Sci Technol 1997;35(5):110. [28] Armstrong J, Armstrong W. Phragmites australisa preliminary study of soil oxidising sites and internal gas transport pathways. New Phytol 1988;108:37382. [29] Brix H. Functions of macrophytes in constructed wetlands. Water Sci Technol 1994;29(4):718. [30] Armstrong W, Armstrong J, Beckett PM. Measurement and modelling of oxygen release from roots of Phragmites australis. In: Cooper PF, Findlater BC, editors. Constructed wetlands in water pollution control. Oxford, UK: Pergamon Presss; 1990. p. 4152. [31] Eriksson PG, Weisner SEB. An experimental study on effects of submersed macropytes on nitrication and

grateful to the Director, MUIENR for material support and Mr. Robert Bikala and Mr. Bright Twesigye for their technical support during system set-up and monitoring.

References
[1] Chale FMM. Plant biomass and nutrient levels of a tropical macrophyte (Cyperus papyrus L.) receiving domestic wastewater. Hydrobiol Bull 1987;21(2):16770. [2] Azza NGT, Kansiime F, Nalubega M, Denny P. Differential permeability of papyrus and Miscanthidium root mats in Nakivubo swamp, Uganda. Aquat Bot 2000;67:16978. [3] Emerton L, Iyango L, Luwum P, Malinga A. The present economic value of Nakivubo urban wetland, Uganda. IUCNThe World Conservation Union, Eastern Africa Regional Ofce, Nairobi and National Wetlands Programme, Wetlands Inspectorate Division, Ministry of Water, Land and Environment, Kampala, 1998. p. 130. [4] Gersberg RM, Elkins BV, Lyon SR, Golman CR. Role of aquatic plants in wastewater treatment by articial wetlands. Water Res 1986;20(3):3638. [5] Reddy KR, Patrick WH, Lindau CW. Nitrication denitrication at the plant rootsediment interface in wetlands. Limnol Oceanogr 1989;34(6):100413. [6] Kansiime F, Nalubega M. Wastewater treatment by a natural tropical wetland: the Nakivubo swamp, Uganda. Processes and implications. Ph.D. thesis, IHE, Delft/ Agricultural University, Wageningen, The Netherlands. Rotterdam, The Netherlands: A.A. Balkema Publishers; 1999. 300pp. [7] Kansiime F, van Bruggen JJA. Distribution and retention of faecal coliforms in the Nakivubo wetland in Kampala, Uganda. Water Sci Technol 2001;44(11/12): 199206. [8] Kipkemboi J, Kansiime F, Denny P. The response of Cyperus papyrus (L.) and Miscanthidium violaceum (K. Schum) Robyns to eutrophication in natural wetlands of Lake Victoria, Uganda. Afr J Aquat Sci 2002;27(1): 1120. [9] American Public Health Association (APHA). Standard methods for the examination of water and waste water, 18th ed. Washington, DC, USA: APHAAWWWA WEF; 1992. [10] Novozamsky I, Houba VJG, van Eck R, van Vark W. A novel digestion technique for multi-element plant analysis. Commun Soil Sci Plant Anal 1983;14(3):23948. [11] Mutulewich VA, Strom PE, Finstein MS. Length of incubation for enumerating nitrifying bacteria present in various environments. Appl Microbiol 1975;29:2658. [12] Schmidt EL, Belser LW. Nitrifying bacteria. In: Page AL, Miller RH, Keeney DR, editors. Methods of soil analysis Part 2: chemical and microbiological properties (2nd). Madison, WI: American Society of Agronomy Monograph, vol. 9, 1982. p. 102742. [13] Belser LW, Mays L. Use of nitrier activity measure ments to estimate the efciency of viable nitrier counts in soils and sediments. Appl Environ Microbiol 1982;43(4): 9458.

ARTICLE IN PRESS
J. Kyambadde et al. / Water Research 38 (2004) 475485 denitrication in ammonium-rich aquatic systems. Limnol Oceanogr 1999;44(8):19939. [32] Heidenwang I, Langheinrich U, Luderitz V. Self-purication in upland and lowland streams. Acta Hydrochim Hydrobiol 2001;29(1):2233. . [33] Korner S. Nitrifying and denitrifying bacteria in epiphytic communities of submerged macrophytes in a treated 485 sewage channel. Acta Hydrochim Hydrobiol 1999;27(1): 2731. [34] Raff J, Hajek P-M. Zur Nitrikation in FlieXgew. ssern a durch suspendierte und Nitrikanten, 1. Mitteilung: Abh. angigkeit der Nitrikationsrate vom C-BSB und Ammoniumgehalt. GWF Gas Wasserfach: Wasser-Abwasser 1981;122:159.